In a move aimed at preserving the integrity and reliability of its content, the English Wikipedia community has officially banned the use of large language models (LLMs) and other AI tools for generating or rewriting article text. The updated guideline, which took effect following a community discussion that closed on March 20, 2026, prohibits editors from relying on artificial intelligence to create or substantially modify the prose of articles.
The decision was reached through a Request for Comment (RfC) process, where the proposal received overwhelming support with 44 votes in favor and only 2 against. This strong consensus led to its swift adoption under Wikipedia’s “SNOW” (snowball) rule for clear community agreement. The new policy replaces earlier, more limited guidelines that primarily restricted AI use for creating entirely new articles from scratch.
According to the updated rules, content generated by AI — including text produced or heavily rewritten by tools such as ChatGPT or similar models — cannot be added to new or existing articles. The primary concerns driving the ban include the risk of “hallucinations” (fabricated information), subtle factual errors that are difficult to detect, and potential biases inherited from training data. Wikipedia’s core principles of verifiability, neutrality, and reliance on reliable sources are seen as incompatible with unchecked AI-generated material, which often lacks proper traceability to original references.
The policy is not an outright rejection of all AI assistance. Two narrow exceptions have been established: editors may use LLMs for basic copyediting tasks (such as improving style, grammar, or clarity) and for assisting with initial translations, provided there is strict human oversight. In all cases, the final responsibility for accuracy, neutrality, and compliance with Wikipedia policies rests entirely with the human editor. AI tools cannot alter the substantive content or introduce new information without direct human verification and sourcing.
This update primarily applies to the English-language edition of Wikipedia, which operates independently from other language versions. Each Wikipedia language community sets its own policies. For instance, the Spanish Wikipedia already maintains restrictions on using LLMs for creating or expanding articles, though its approach may differ in specifics regarding copyediting and translation support.
Editors are reminded that any persistent violation of these guidelines could be treated as disruptive editing, potentially leading to blocks from further contributions.
Discussion about this post